Dear Antanas,
I still think this source is very likely to be related to eye blinks.
0.5-40Hz with baseline correction will not remove eye blinks and if
they are not tightly time locked to the stimulus (which is probably
the case) they won't generate a clear peak but a low-frequency
component that can still be quite strong and affect the inverse
solution. I think you shouldn't get too fixated on the maximal
activation. This was just a heuristic to display something meaningful
for solution which is basically a movie. Sometimes it works sometimes
it doesn't. The really important question is what sources show up in
your contrasts of interest and for that you should do proper
statistical analysis as we discussed previously.
Best,
Vladimir
On Thu, Aug 27, 2009 at 5:17 PM, Spokas
Antanas<[log in to unmask]> wrote:
> Dear Vladimir,
>
> Thank you so much for your answers, I've learned quite a lot from your
> emails!
> But I am not sure about the eye blink, becouse the peak I am refering to, is
> not in my time series, (as I've mentioned there is no activity in my frontal
> electrodes but only posterior one's indicating N1 and P1 visual
> processing) but it's in the results of SPM source reconstruction where it
> gives the time course of the estimated most active sources over time. And
> that's where this peak is displayed as the most active source activation
> peak over the whole time series and over the whole brain and its the only
> peak of that estimated source activity? I have as well bandpassed my data
> with 0.5-40Hz and removed baseline average from it so there is no high
> frequency noise and quite clear ERPs. Thank you.
>
> Regards,
> Antanas
>
> ________________________________
> From: Vladimir Litvak <[log in to unmask]>
> To: [log in to unmask]
> Sent: Thursday, 27 August, 2009 16:43:39
> Subject: Re: [SPM] imaging source reconstruction
>
> Dear Antanas,
>
> On Thu, Aug 27, 2009 at 2:52 PM, Spokas
> Antanas<[log in to unmask]> wrote:
>> Thank you for your answer, you're right that predicted and observed scalp
>> maps match at that time point, it's just that in the region there is
>> negative activation and source activation is shown high positive
>> activation,
>
> This is quite normal. The relation between source polarity and scalp
> polarity depends on the orientation of the source. You are used to
> sources oriented upwards but this source as you are saying is on the
> inferior surface of the frontal cortex (but see below) so it is
> oriented downwards and when it has positive activation what you see on
> the top of the head is negative. I again refer you to the dipole
> simulator from my previous e-mail.
>
>> and it's quite a sharp peak lasting for very short time in that source
>> time
>> course and it is hard for me to interpret that activation, as it's at
>> 100ms
>> poststimulus and in the averaged time series and this activated source is
>> in
>> the inferior frontal cortex, when I would only expect occipital
>> activation?
>
> When I hear something like this I immediately think 'eyeblink'. It's
> indeed unlikely that you will see true orbitofrontal source but it is
> very likely that your data is contaminated with eye-blinks and since
> your subjects were probably consistently blinking around the stimulus
> presentation these eyeblinks are not averaged out. Since SPM does not
> model sources in the eyes, this activation is explained by the closest
> source on the mesh which is orbitofrontal cortex. There are several
> things you can do ranging: just ignore this source with the hope that
> SPM will still model the occipital brain sources correctly, go back to
> your single trials and reject the ones with eyeblinks, high pass
> filter your data hoping to get rid of at least some it etc. In general
> the best solution is to instruct your subjects to fixate and not blink
> during stimulus presentation and then reject all the trials with
> blinks.
>
>> Another thing is that in ERP literature it's quite usual to find direct
>> reference from active electrodes to the brain area activated, e.g.
>> occipital
>> electrodes will be indicating occipital cortex involvment...? Thank you.
>>
>
> This is perhaps true for the occipital cortex where because of the
> anatomy the brain sources project to the back of the head, but not
> true in general.
>
> Best,
>
> Vladimir
>
>> ________________________________
>> From: Vladimir Litvak <[log in to unmask]>
>> To: [log in to unmask]
>> Sent: Thursday, 27 August, 2009 11:30:13
>> Subject: Re: [SPM] imaging source reconstruction
>>
>> Dear Antanas,
>>
>> The relation between where the source is and where the electrodes
>> record the maximal deflection is not a simple one (you can play with
>> the tool from
>>
>> ftp://www.besa.de/be/besa.de/free_tools/DipoleSimulator-2009-03-27-Install.exe
>> to get some intuition). Did you try to compare in the rendering tool
>> the predicted and the observed scalp maps? If they match then perhaps
>> SPM does a good job after all. If not then I can look at the example
>> and tell you if there is something wrong. Also remember that SPM does
>> not reconstruct each time frame separately (like some other packages)
>> but looks at temporal and spatial modes computed over the whole time
>> window.
>>
>> Best,
>>
>> Vladimir
>>
>> On Thu, Aug 27, 2009 at 10:21 AM, Spokas
>> Antanas<[log in to unmask]> wrote:
>>> Dear Vladimir,
>>>
>>> I've incounted troubles with imaging source reconstruction, I've just
>>> noticed, that at some point in time series, it's estimated source with
>>> highest activation over the whole brain in the region, where none of the
>>> electrodes have recorded much of a potential and completely ignored much
>>> higher potential at different sight?.. meaning that the whole estimation
>>> flies out of the window? I've checked again all electrode positions in
>>> cooregistration, and all looks good there. Could you please explain
>>> Vladimir, thank you.
>>>
>>> Regards,
>>> Antanas
>>>
>>>
>>>
>>
>>
>
>
|